The Introduction of Writing into Achaemenid Gandhāra and the Origin of Pāṇini’s Grammar

Abstract

In this article, I argue that the composition of Pāṇini’s Aṣṭādhyāyī (fourth century bce), a sophisticated derivational grammar of Sanskrit, was made possible by the introduction of writing into ancient India in the wake of the Achaemenids’ conquest of Gandhāra in the late sixth century bce. Various theoretical and terminological features of Pāṇini’s grammar can be adduced as evidence that it was produced in a literate environment. Here I present and discuss some of the arguments against Pāṇini’s literacy, pointing out their weaknesses, in particular their underestimation of the radical novelty of Pāṇini’s work, which marked an epochal shift towards the secularisation of knowledge, and reliance on the myth of the powerful orality.

Encounters with different peoples and cultures have often marked turning points in the history of civilisations, since they entail the exposure to new ideas, technologies, religious beliefs, art forms, lifestyles, and so on. One such encounter in the history of ancient India was that with the Persian civilisation of the Achaemenid Empire, ¹ which annexed some of the north-western regions of historical India in the late sixth century bce. Shortly after that, sometime in the second half of the first millennium bce, two major events in Indian cultural and social history took place in one of those regions known at the time as Gandhāra, which roughly corresponds to the area of today’s northern Pakistan around Peshawar and Islamabad. First, writing was introduced to the subcontinent. ² Second, the grammarian Pāṇini composed his work, the Aṣṭādhyāyī, an accurate, exhaustive and formal description of a high variety of the language later known as Sanskrit, which became the universally accepted standard of ‘correct’ language throughout its long history.

Is there a connection between these two events? I attempted to answer this question in an earlier publication, ³ where I concluded that the introduction of writing was the crucial element that allowed the Brahmanical civilisation to take the momentous ‘cognitive leap’ making the composition of the Aṣṭādhyāyī possible. Here I will develop some considerations on the arguments that have led me to draw this conclusion and try to respond to some generic critique from the champions of orality.

A Brief Survey of the Early History of Writing in India

Neither of the two events mentioned above can be dated with absolute certainty. Based on textual evidence, some Indianists—among them, most prominently Bühler ⁴ —have argued that a literate culture can already be detected in the works of the late Vedic period (first half of the first millennium bce), but as Salomon notes, ‘the absence of a single explicit and indisputable reference to writing anywhere in early Vedic literature suggests that the Vedic culture was a preliterate one’. ⁵

It is only around the mid-third century bce that writing is first attested in India in the inscriptions of the Maurya emperor Aśoka. Most Aśokan inscriptions are written in Brāhmī, the progenitor of all later Indian scripts, but two rock edicts found in eastern Gandhāra, at Shāhbazgaṛhī and at Mansehrā, are written in another early Indian script, Kharoœṭhī, ⁶ which was in use in the region and the surrounding areas. Kharoœṭhī was mostly, though not exclusively, used for Buddhist texts in Gāndhārī, the local Prakrit, that is, Middle Indo-Aryan vernacular. ⁷ It was used for a few centuries in the areas where the rock edicts were found, and even spread to Central Asia through the trans-regional networks of Buddhist religious establishments, but eventually died out in the early centuries ce.

Regarding the Kharoœṭhī of the Aśokan inscriptions, ⁸ Salomon notes that

the script appears as a more or less fully developed system, which may be taken to indicate a significant prehistory. But it is difficult to estimate how long such a presumed developmental period might have taken, though it is likely on historical and palaeographic grounds that Kharoœṭhī originated some time during the Achaemenian era. ⁹

It should also be noted that the language of most Aśokan inscriptions ¹⁰ is a variety of Prakrit, whereas the earliest epigraphical attestations of Sanskrit are considerably more recent, dating to around the beginning of the Common Era. ¹¹ The relationship between Brāhmī, which is written from left to right, and Kharoœṭhī, which is written right to left, has been the object of much controversy, but today most scholars agree that Kharoœṭhī is older and may have had an influence on the development of Brāhmī ¹² since, despite the apparent differences, the two systems share the same basic structural principles. In particular, consonantal signs are to be read in both scripts with an inherent vowel a, while the other vowels (ā, i/ī, etc.) are noted by means of diacritic strokes to the top or bottom of the consonantal grapheme or by independent signs when they occur in syllable-initial position.

The Place and Date of Pāṇini

The Indian tradition maintains that Pāṇini hailed from Śalātura, a village in Gandhāra. ¹³ Modern scholarship has generally accepted this view. Various authors have noted that, linguistically, the language described by Pāṇini is a late Vedic north-western variety of Sanskrit. Thieme writes:

Nothing can confirm this tradition more conclusively that Pāṇini’s relation to the Black Y[ajur]V[eda]. As a northern grammarian he is familiar with the north-western vedic schools and ignores the White Y[ajur]V[eda]. Even the Bhaœā [i.e., the everyday language] as taught in his grammar has some peculiar forms in common with the K[āṭhaka]S[aṃhitā] …which must be northern idioms. ¹⁴

Deshpande (1983) discusses some of Pāṇini’s rules concerning certain sandhis and the syntax of infinitives (to some extent different from those found in classical Sanskrit) and observes that they seem to reflect north-western usages, calling him a ‘frontier grammarian’. ¹⁵

The geographical references in the Aṣṭādhyāyī itself also corroborate the tradition. Renou, while denying the ‘north-western’ character of Pāṇini’s language, notes nevertheless that ‘such minute details of toponymy are given for these regions as are not found for others’. ¹⁶ Thus, Pāṇini himself mentions Śalātura in a sūtra and the famous town of Taxila (Takœaœīlā in Sanskrit), situated nearby, in another. ¹⁷ The name Gandhāra also appears, indirectly, in the term gāndhāri (‘a nobleman from Gandhāra’) in A.4.1.169. Several other names of regions, mountains, rivers, towns and villages in the north-west part of the subcontinent are also mentioned, ¹⁸ but only a few from further east and south. ¹⁹ Patañjali (ca. 150 bce), the author of the Mahābhāṣya, the earliest known commentary on the Aṣṭādhyāyī, quotes a verse under A.1.1.20 calling Pāṇini Dākœīputra, the son of Dākœī, ‘after the name of his mother who was of the Dakœa gotra’. ²⁰

Pāṇini is generally dated to around 350 bce, ²¹ based on contextual evidence. ²² Cardona himself, the doyen of Pāṇinian studies, concluded in 1976 for a date between the early to mid-fourth century bce, ²³ but in a recent article he appears to favour an even earlier date, around 500 bce ‘in view of the state of the language that Pāṇini describes’. ²⁴

That Pāṇini knew of writing is pretty much certain since sūtra A.3.2.21 teaches how to derive the compound lipikāra ‘scribe’, which Salomon describes as ‘the most cogent single piece of literary evidence for writing before the Mauryan period’, remarking that there is ‘no cogent reason to rule out’ that the script in question may be Kharoœṭhī (see also Norman). ²⁵ Others have argued that it might be Aramaic, since no evidence of pre-Aśokan Kharoœṭhī has been found yet. ²⁶ And of course, it is very possible that Pāṇini was aware of the existence of writing (and of professional scribes) through his exposure to Persian (and possibly Greek) culture, but he was not himself literate.

In conclusion, the historical evidence, based on textual and epigraphical sources, does not allow us to affirm with absolute certainty that Pāṇini knew writing nor to exclude it categorically. This is why in my previous article on this topic I have argued that we should rather focus on the Aœṭādhyāyī itself and assess its conceptual, structural and terminological features to decide whether it might or might not have been produced in an illiterate environment.

The Scenario for Pāṇini’s Literacy

Salomon describes Gandhāra as ‘an archetypal frontier region situated on a fault line between major geo-cultural zones’, namely, ‘the Iranian world to the west, the Indian world to the east, and Central Asia to the north’. And he notes that as a ‘zone of transit between several cultural regions’, it ‘has always had a distinct and complex cultural identity’ but that it was ‘more part of India than of the Iranian world, despite the constant influence of and frequent political domination from the west’. ²⁷

It is reasonable to assume that, after Gandhāra was annexed by the Achaemenid Empire, the contacts with the Persian culture intensified. However, Petrie and Magee note that ‘the Achaemenids appear to have made use of local kings in the easternmost regions, rather than appointing satraps of their own’, and that ‘there is minimal evidence for direct Achaemenid control, and good evidence for the maintenance of local authority’. ²⁸ Nevertheless, it is generally accepted that Kharoœṭhī originated as an adaptation of the Aramaic script to an Indian language (possibly Gāndhārī, the local vernacular); moreover, the above-mentioned Aśokan inscriptions in Aramaic found in the region also suggest that, almost a century after the end of the Achaemenid rule, ²⁹ the memory of the importance of the empire’s administrative language and script was still alive (and, perhaps, there were still a few people around who could read it).

Thus, it appears that Pāṇini was active in a diverse multicultural and plurilingual intellectual environment, but somewhat peripheral and cut off from the heartland of Brahmanism, by then identified with the Gangetic plain. ³⁰ As I have argued before, these circumstances, revealing and exacerbating the vulnerability of the Brahmanical identity in a changing world, might have prompted the efforts of a few generations of grammarians—culminating in Pāṇini’s opus magnum—to produce a full grammatical description of the language that was central to that identity, the tongue of the Āryas that centuries later (largely thanks to their endeavour) came to be known as saṃskṛta, the perfected language, refined and regulated by Pāṇini’s grammar.

I believe that Pāṇini and his unknown predecessors ³¹ could not have succeeded in their enterprise without having access to the newly introduced technology of writing. Their feat required what I have called a ‘cognitive leap’ because I believe that a grammar such as Pāṇini’s—possibly any full grammatical description of a language that is not exclusively literary and confined to a closed corpus—could not even have been conceived in a primary oral culture, that is, a culture with no practice of writing. ³²

This conclusion is controversial—much old and recent scholarship would object to it, as I discuss below—and needs to be bolstered by evidence, which can only be of an inferential and circumstantial nature. It is unlikely that archaeological proof will ever emerge showing that Pāṇini and his contemporaries knew and used writing. ³³ Instead of pursuing this course, therefore, I have looked for evidence of the key role that literacy played in the emergence of the classical grammatical tradition inside Pāṇini’s work. But first, let us look at the arguments of those who claim that his work was an oral composition.

Some Views of the Advocates of the Oral Composition of the Aṣṭādhyāyī

The arguments of those who maintain that the Aṣṭādhyāyī was composed orally can be summarised under the following headings:

The lack of hard evidence (such as inscriptions or manuscripts) attesting the use of writing in India before Aśoka’s time;

The inadequacy of the scripts (Aramaic or Kharoœṭhī) that Pāṇini might have known and used to reproduce Sanskrit;

The insistence on the primacy of oral teaching and transmission in the ancient Indian context;

The very limited references to writing in the Aṣṭādhyāyī.

As for the first point, despite the said lack of evidence, there is general scholarly consensus, as shown above, that Pāṇini was active in a time and place where the local population must have been aware of the existence and purposes of writing because of the Aramaic script used by the Achaemenid administration. The fact that his grammar has a rule for the derivation of the term lipikāra ‘scribe’ proves that at the very least he knew of writing. Thus, it cannot be excluded that he also knew how to write, namely, that he was literate.

Regarding the second point, the inadequacy of the scripts to write Sanskrit, many of the arguments adduced as evidence can be shown to be weak or questionable, as I argue below. Most importantly, they appear to be based on a fundamentally misplaced assumption, namely, that the primary purpose of writing is the representation of the spoken language rather than being a medium to communicate facts, ideas and feelings. ³⁴ If the goal is to communicate, it is not indispensable or even desirable to represent the underlying language as accurately as possible, as the spelling of English and French unquestionably shows.

Thirdly, it seems to me that the emphasis on the orality of teaching and transmission, largely justified by the testimony of the textual sources and the consideration of the cultural environment in which Pāṇini’s grammar was born and thrived, is on the whole irrelevant to the issue of its conception and composition, although it can help to explain certain aspects of its genesis, as I argue below.

As for the absence of clear references to writing and so on (besides the term lipikāra), it is worth keeping in mind that the Aṣṭādhyāyī is a collection of rules for the derivation of words, not a lexicon, and therefore the occurrence of lexical items largely depends on the regularity/irregularity of their formation and related derivatives. Thus, a hypothetical Pāṇinian grammar of English would certainly contain the word ‘child’ because the plural ‘children’ is irregular, but not necessarily the words ‘infant’, ‘kid’, ‘lad’, ‘boy’, ‘girl’, etc., since they form their plural according to a general rule (the addition of the suffix -s). ³⁵ As for the lexical items that do appear in the Aṣṭādhyāyī, it is nearly impossible to determine with absolute certainty what they refer to due to the absence of a meaningful narrative or explicative context. ³⁶ Therefore, many conclusions reached on the basis of Pāṇini’s lexicon appear to be far too bold and highly subjective. For example, in an article dealing with foreign terms in Sanskrit pertaining to writing, Falk writes that Pāṇini presents the word for ‘script’ in A.3.2.21 ‘in two forms, lipi and libi, without hinting at which he preferred. This means that he regarded this word as outside his own science … .’ ³⁷ I would counter that the sūtra simply records the fact that both variants of the word were in use in Pāṇini’s time, and neither was prevalent; as loanwords that could not be derived from a Sanskrit verbal root, their grammaticality could not be assessed. In fact, their very mention in the sūtra shows that he considered the term to be within the scope of his science.

Scholars who are sceptical about Pāṇini’s reliance on writing for the composition of the Aṣṭādhyāyī generally point to the strong ‘memory culture’ ³⁸ of ancient and medieval India. They maintain that the oral transmission of the Aṣṭādhyāyī, a relatively short text that can be entirely recited in a couple of hours, presented no challenge to a culture where certain specialists were used to memorising thousands of texts, both in verse and in prose, thanks to the zealous cultivation of the arts of memory. ³⁹ But, as Bronkhorst remarks:

the oral composition of a work as complex as Pānini’s grammar is not only without parallel in other human cultures, it is without parallel in India itself. [ … ] the least one can ask for is some kind of indication as to how Panini did it. ⁴⁰

Thus, Falk writes:

Pāṇini … did not use writing for his grammar, which was based solely on memory and the human faculty to associate related rules. His grammar presupposes an elaborate and absolutely encompassing system of sound analysis, with names for each sound and precise rules for the changes sounds undergo when juxtaposed in a word or sentence. ⁴¹

But, in this way, he eschews any considerations of how Pāṇini might have composed a grammatical work of such sophistication and complexity by purely oral means.

Aware of the complexity of the grammar, Staal insists that it was composed orally but conjectures that Pāṇini might have relied on the help of junior colleagues or pupils who memorised different parts of the work in progress and recalled them at the teacher’s request. ⁴² Staal’s hypothetical scenario is based to some extent on his observation of the recitational practices for Vedic texts among contemporary Brahmans in Kerala, ⁴³ which were of exceptional complexity and—at the time, in the 1950s—still enjoyed considerable social prestige. In any case, it could possibly explain some organisational aspects of the process of composition, but it does not even attempt to articulate a broader explanation of how the conception of a complex derivational grammar might have come about in a purely oral culture and developed a sophisticated methodology and metalanguage. ⁴⁴

Scharfe concedes that:

[w]e might speculate that [Pāṇini] used such writing, inadequate as it was, to help in organizing his material; but it is hard to imagine that his grammar could have been written down adequately. The grammar was passed on orally, with pitch accents and nasalized vowel tags (along with consonantal tags) added on as markers. [ … ] None of these articulative features have survived in our text of the Aṣṭādhyāyī. If a written form of the grammar was handed down along with the oral transmission, it would have played a secondary role in backing up the student’s memory. ⁴⁵

Similarly, Witzel, while accepting that Pāṇini was probably aware of writing, maintains that he composed and transmitted his work orally. ⁴⁶ However, his arguments—like Scharfe’s—essentially rest on the perceived inadequacies of Kharoœṭhī to represent Sanskrit. They gloss over the challenge the grammar posed at the conceptual and compositional levels and focus instead on formal aspects of the script, whose shortcomings might indeed have been problematic if the circulation and transmission of the grammar had mainly relied on the written text, without the accompanying support of oral teaching—an unlikely scenario, as I have argued elsewhere.

Claiming to ‘follow in the footsteps of Staal’, in 1990, ⁴⁷ Falk launched a justified attack on the rash conclusions reached by Goody in 1987, in a chapter devoted to the transmission of the Vedas in ancient India. ⁴⁸ Falk lamented that Goody’s argument rests on outdated views ultimately going back to Bühler. ⁴⁹ He noted that Goody’s conclusion that there is ‘little doubt that Pāṇini used writing to enable him to formulate the “rules” of grammar’ was based on some features of the Aṣṭādhyāyī such as ‘the existence of a table of array of the sounds, the use of pratyāhāras, and the lists of verbal roots’. ⁵⁰ Then Falk remarks:

To say that some feature is met with in literate societies does not mean that it must belong exclusively to this kind of culture. Seen the other way round, we can say that none of Pāṇini’s devices … recurs in any of the literate cultures of antiquity to a comparable extent.

Based on these observations, he concludes:

The degree of abstract thinking behind Pāṇini’s system is very high compared to other contemporaneous cultures in any part of the world. In fact it is difficult for us to imagine today how such an intricate system could have been developed without writing. But this is our fault and not Pāṇini’s. Before Pāṇini perfected the system there were many generations in different parts of the subcontinent working on it and it is impossible to reconstruct the steps or to estimate the span of time needed to lead to such an end. ⁵¹

As noted by Bronkhorst, this line of reasoning does not make a particularly robust argument. ⁵² One can legitimately expect Falk and others who share his views to offer at least a tentative explanation of how Pāṇini could have produced his work in a purely oral context.

Falk admits that the Aṣṭādhyāyī is a unique example of systematic, abstract thinking about language in the ancient world. As already noted by many scholars, ⁵³ Pāṇini’s grammar is not a descriptive one, but rather a derivational grammar that operates with abstract elements (bases and affixes) and provides rules for their combination into fully formed words and, ultimately, sentences. Auroux has pointed out that proper linguistic knowledge, which moves beyond the ‘epilinguistic’ knowledge that is probably found in any human culture, only appears in literate societies because it requires a level of objectification of language that writing alone can provide. ⁵⁴ When a culture becomes literate, its language ceases to live exclusively in the ephemeral verbal exchanges of the members of the speaking community and acquires an external existence and a stability that facilitate the observation and description of its features and structures. To put it differently, writing allows the transition from epilinguistic to metalinguistic knowledge. Olson has argued that, in fact, while historically writing has usually developed as a tool for pragmatic purposes (trade, accountancy, administration, etc.), it inevitably entails a reflection on language because it requires its inventors to make choices about what aspects of speech it should reproduce and how. ⁵⁵ In that sense, writing is itself a metalinguistic tool, leading to the segmentation of the spoken chain into its perceived components, such as sentences, words and sounds.

Perhaps, what makes ancient Indian culture exceptional is the fact that it apparently began this analytical process well before the introduction of writing as a consequence of its marked concern for the transmission of the sacred texts, the Vedas, in pristine form. The accurate memorisation of texts (usually, one recension of a given Vedic Saṃhitā and its related exegetical texts) was the collective effort of ‘schools’ (caraṇa) whose members were the male Brahmans of specific lineages, who regarded their task as a religious and social duty. My hypothesis is that, already at the pre-literate stage of Vedic culture, this joint effort—consisting of teaching the texts to students, rehearsing them in private or in public, performing their recitation before learned audiences or during religious ceremonies—supported by advanced mnemonic techniques, conferred a fixed, stable textuality to the works being transmitted, which were, thus, shielded from the unpredictable variability of daily speech, and consequently experienced an existence independent of single individual speakers. This may have paved the way to fine and accurate observations of the properties of the language—phonological, syntactical, lexical, and so on—in parallel with the invention of alternative modes of recitation (the krama, jaṭā and ghana) that entailed cutting up the text into segments and, eventually, led to the composition of the padapaṭhas, in which a Vedic saṃhitā is analysed into words (pada), and of prātiśākhyas, works which provide the rules to reconstitute the continuous text (saṃhitāpāṭha) from the padapaṭha and also deal with general phonetic matters. ⁵⁶

This phase of grammar must have originated and developed during the oral phase of Vedic culture, in all likelihood. ⁵⁷ But I think it contributed to creating the intellectual background (consisting of phonological knowledge and proto-grammatical analysis as well as the conceptual apparatus and methodology of ritualism) that allowed the Brahmans of north-west India, once they became acquainted with writing through prolonged contact with the literate culture of Achaemenid Persia, to put the new technology to good use and make huge and very rapid progress in their linguistic knowledge, culminating—in roughly one century and a half, if the date around 350 bce for Pāṇini is correct—in the composition of the Aṣṭādhyāyī.

When I call the progress made by Pāṇini’s anonymous predecessors ‘huge and very rapid’, I have in mind an aspect of Pāṇinian grammar that the advocates of the ‘orality view’ underestimate or simply ignore, namely, the fact that it is not just a grammar of the Vedic language, which is in fact treated in a quite cursory way in the Aṣṭādhyāyī, but of the whole ordinary language, ⁵⁸ which did not exist only in literary corpora the oral transmission of which was ensured by a whole class of dedicated specialists, but lived instead in countless verbal exchanges in the many diverse contexts of ordinary life. As I have already pointed out in my previous article, the abstract elements—the bases and affixes—with which Pāṇini’s system operates pre-exist it; they are a given. We can suppose that, thanks to the new possibilities opened up by literacy, some Vedic grammarians, who had traditionally focused on the task of the preservation of the Vedic language handed down in the sacred texts, gradually expanded their investigation to the worldly language, leading to the formidable effort of analysis presupposed by the Aṣṭādhyāyī. I think that in a primary oral context, this kind of minute, painstaking, extensive work of observation, collection of data and systematic classification—presupposed by the Aṣṭādhyāyī—would have been impossible, both conceptually and practically. ⁵⁹ As Olson notes, ‘cultures have evolved and accumulated both practices, concepts and categories for thinking about language, the world and the self. These practices and associated concepts are what give thinking a history’. If literacy did play such a role in ancient India, we should expect to find ‘the evolution of a new set of concepts, concepts brought into focus in the attempts to use writing for a variety of social purposes such as science, law, literature, history, philosophy and religion’. ⁶⁰

My impression is that those who uphold the view that Pāṇini composed his grammar without any recourse to writing dismiss too lightly a number of aspects related to the nature of his amazing work and its genesis. They lump the Aṣṭādhyāyī together with Vedic literature, neglecting the fact that Pāṇini’s grammar had no value for the preservation of the Vedas and the performance of the related ritual (as shown by the Mīmāṃsakas’ critique levelled at it centuries later). They also ignore the momentous implications of the shift that allowed the early grammarians to direct their attention for the first time to a worldly phenomenon, such as the ordinary language, with no apparent connections to the sphere of ritual (and therefore outside the scope of the orthodox Vedāṅgas), and consider it a worthwhile object of investigation and study. I argue that this shift was made possible by the introduction of writing. Those who think otherwise should offer an answer to some of the anything but banal questions: how did this change of attitude come about? What were the historical and social conditions that led to it? How should we explain, for example, the conceptual transition from the method of morphophonemic change followed in the Prātisākhyas to that of substitution (ādeśa) ⁶¹ adopted by Pāṇini? If we are to accept that Pāṇini did not rely on writing, we need a credible narrative that can explain this radical epistemic shift.

Furthermore, the insistence on the oral transmission and teaching of Pāṇinian grammar (which no one actually contests) glosses over the fact that its preservation and circulation have hardly anything to do with its conception and composition. The upholders of the orality view respond to the remark that anywhere else in the world linguistic knowledge presupposes writing, by appealing to the exceptionality of ancient India in that respect. Most remarkably, they invoke at every step the powerful ‘memory culture’ (to borrow a phrase used in Houben and Rath ⁶² ) of ancient India, understandably impressed by the astonishing mnemonic skills exhibited—even today—by traditionally educated Sanskrit scholars in India, skills that in ancient times ensured the oral preservation of a substantial corpus of Vedic texts, both in verse and prose, for hundreds of years. When questioned (admittedly, in Goody’s poor arguments based on shaky, second-hand knowledge of the ancient Indian context) about the plausibility from a cognitive point of view of the historical scenario in which Pāṇini would have composed his highly abstract and formalistic derivational grammar orally, Falk (quoted above) responds that it is our limitation if we cannot figure out how Pāṇini managed to do so. ⁶³ Presumably, he means that we are too steeped in a literate culture to put ourselves in the shoes of a genius thinker who allegedly lived in a non-literate culture. But such an answer has a fundamental flaw: it does not take into account that those who cultivate the myth of the powerful orality are as irremediably distant from the experience and sensibility of the oral mind as the rest of us. The allegation can be legitimately turned against them.

To conclude with Olson’s words:

Modern readers immersed in a reading culture and equipped with a literate consciousness may find it difficult to recover an ‘innocent eye’, an eye untainted by the experience of reading and writing. Reconstructing a world without words, letters, sentences and documents cannot be reached by introspection alone but is a task for research and theory. ⁶⁴ [my emphasis, VV]

This wise reminder holds true, I think, for anyone who sets out to prove or disprove that Pāṇini was literate.